Compute Poisson deviances (up to a constant) for given design matrix

Share:

Description

A function called within QL.fit to compute Poisson deviances of each gene for given design matrix

Usage

1
2
     PoisDev(counts,design,log.offset,print.progress=TRUE)
     

Arguments

counts

RNA-seq data matrix of integer expression counts. Each row contains observations from a single gene. Each column contains observations from a single experimental unit.

design

A single element from the design.list argument given to QL.fit.

log.offset

A vector of log-scale, additive factors used to adjust estimated log-scale means for differences in library sizes across samples. Commonly used offsets include,log.offset=log(colSums(counts)) or log.offset=log(apply(counts,2,quantile,.75)). The default setting in QL.fit makes no adjustment for library sizes (i.e. log.offset=0).

print.progress

logical. If TRUE, the function will provide an update on what gene number is being analyzed. Updates occur frequently to start then eventually occur every 5000 genes. Note that updates will not occur for one-factor designs, for which closed form solutions are available.

Value

list containing:

dev

vector containing the deviance for each gene under a Poisson model fit to design matrix (or vector, for one-factor experiments) specified by design. This vector is passed along within the QL.fit function.

means

matrix of fitted mean values for each gene

parms

matrix of estimated coefficients for each gene

Author(s)

Steve Lund lundsp@gmail.com